Panel 10: Prompted By: Insights on Generative AI Creation
Chair: Kate Pullinger
10.01 – Infinite Eddies and Infinity +1: Experiments in Elegy co-creating with Midjourney and Twine
Siobhan O’Flynn (University of Toronto)
The first project, Infinity +1 (2022), experiments with the aesthetic and affective possibilities of the story-game platform, Twine, pairing text with images co-created with Midjourney. The project’s genesis began in hearing my father’s fluid movements through time and place in his last months with age-related memory loss and early dementia. Working with the affordances of interactive media, Infinity +1 evokes a sense of the multivocality of identity, personhood, and memory, and the associative movement between persons, places, stories, and interactions. What this multivalence over time revealed was the compensatory dimension of my father’s immersion in other times, and a shifting “personhood” (Higgs 2015) in diminished connection to the present, that Infinity +1 extends.
Infinite Eddies (2023) explores the affective dimension of Midjourney’s generative AI mediating one source photograph of my father and myself. Prompts recontextualize our mediated image in other places, times, narrative realms, and visual aesthetics, yet across the set the configuration of our bodies, our expressions, remain constants. I theorize this manifest recurrence in dialogue with what theorist Hiroki Azuma calls anime’s “moe-elements,” the tagged visual or other elements searchable via the otaku engine, TINAMI (eg. mermaid; tail). Where “moe-elements” detach from the anime character in recombinatory new formations, Midjourney’s cumulative reinscriptions of the physical configuration of our bodies suggests an alternative “affective element” (Kalinovski 2022) in the visual evocation of a relationship, expressed in myriad variations and mutations. Playing with prompts to influence the computational temperature of results, this recurrent, mutable element troubles Benjamin’s formulation of the aura of the original work of art, conjuring instead an aura of affect in social relation and physical configuration, retained across an expanding set of images that are not photographs. The aggregate effect is of eddies or ripples around the source photo, the unseen pebble dropped in a virtual pool.
10.02 – Synthetic Theory for Synthetic Media
Tom Livingstone (UWE)
This paper will re-appraise the work of Vilem Flusser to explore whether or not his media theory can offer any critical strategies to engage the wave of synthetic media growing across text, image, sound and video.
Interest in Flusser’s work has been driven by the prescience of his thought and the applicability of his key concepts to digital media, culture and epistemology. However, synthetic media presents a challenge to Flusser’s media-topology with its strong demarcation of the structural differences between text and image. I will argue that synthetic media’s logic of probabilistic and adversarial generation from vectorized databases, means that distinguishing between text and image is no longer a functional approach to synthetic media objects: AI images are atemporal signifiers without signifieds, AI texts don’t reflect linear thinking.
Furthermore, synthetic media are increasingly augmenting everyday media experience. Beyond ChatGPT, media apparatuses from game engines to smart speakers, boast a growing range of AI-enhanced functionalities. It is becoming increasingly difficult to differentiate synthetic media objects from their formal predecessors (i.e. photographs/customer service responses). As a result, the underlying structures of synthetic media are becoming embedded in everyday life, causing what Flusser might term a media-epistemological revolution. I’ll illustrate this exploration with a set of representative synthetic media objects – such as AI “photography” – that invite and defy straightforward modes of perception.
Within this context, I’ll consider how synthetic media, with their structural horizontalization of text, image, sound, and video, might intersect with storytelling practices. Flusser advocated for a criticism that playfully engaged with the programme of an apparatus to generate informative (as opposed to redundant) media. I’ll conclude by thinking about what a playful use of synthetic media and its multi-modal outputs, bypassing the formal cues of texts and images, might look like.
10.03 – Tectonic – an AR poem-scape
Caitlin Fisher (Immersive Storytelling Lab, York University)
“Feel the Ghosts Hanging Around in the Shape of Love – an AR poem-scape
This artist talk will walk through and discuss the research-creation process of making a room scale AR poem-scape. The text was computationally generated based on an archive of poems written by the author and the AR environment was created using the AI image generator Midjourney and off-the-shelf software (the piece is in two versions – one built for WebXR using 8th Wall and one built in Adobe Aero). The piece is a jumping off point to discuss the power and limits of co-creation with machines, combinatory poetics, and ways to think about the relationship between the text and the immersive navigable environment, including the creation of user pathways, sound placement, text as image versus text as spoken word, and the translation of a linear, computationally-generated poem for a spatial environment. The talk ends on a speculative note – what can this poem-scape and work like it, built on the cusp of AI + XR creation, suggest about future forms?
The girl’s body soft as oatmeal
And smelling like opium
at lunch I go to her with the eyes of a car.
diamonds move within us Ancient. new Look: we made this.
This potent Deep dive into two people
fold enough of them together for a long staircase
Narrow and dirty
might as well be fucking Ithaca.
standstillstandstillstandstill hands together, fragile, Each other’s most beautiful and impossible pressure
Look: we made this.
Reach out, feel the ghosts hanging around in The shape of love
A clear golden shot from my soul to your bed, screaming your name.